Department of
Applied
Mathematics

Algorithmic Game Theory and Poker (Jirka Matoušek Prize talk)

Martin Schmid

Abstract

Solving games has always been a major scientific challenge and the successful stories are well-known historical milestones.

1996/1997 Garry Kasparov lost to IBM Deep Blue

2007 University of Alberta Checkers unbeatable

2015 University of Alberta Limit-Holdem Poker unbeatable

2015/2016 Google DeepMind Go: professional player lost to computer program.

In this lecture, I will introduce Poker as a major challenge in artificial intelligence and explain the core mathematical and algorithmic properties. We will see that Poker is not only more challenging game than Go or Chess, but that it also promises more appealing real-world applications. After the lecture, we will understand why despite great theoretical and algorithmic advances in recent years, computers are still unable to beat humans in this challenging game.